Effective intensity-based 2D/3D rigid registration between fluoroscopic X-ray and CT

D. Knaan and L. Joskowicz

School of Computer Science and Engineering The Hebrew University of
Jerusalem, Jerusalem 91904, Israel.

Abstract:

We present an effective intensity-based method for rigid registration of a
patient preoperative CT to its intraoperative situation with a few
fluoroscopic X-ray images obtained with a tracked C-arm.  It improves upon
existing methods and overcomes most of their intrinsic speed, robustness,
and accuracy problems.  For speed, we generate Digitally Reconstructed
Radiographs on small, dynamically selected regions of interest from
precomputed ray gray levels in expected viewing directions, and use a
multiresolution hierarchy of fluoroscopic X-ray images.  For robustness and
accuracy, we use a two-step comparison measure: Normalized Cross
Correlation followed by Variance Weighted Sum of Local Normalized
Correlation.  To avoid local minima, we use a genetic search method.  Our
experiments on simulated, in-vitro, and cadaver data show an overall mean
Target Registration Error of 0.8mm (max=1.5mm), 95\% of the time, computed
in 20--100 seconds under realistic conditions.

Keywords: 2D/3D rigid registration, X-ray to CT registration,
intensity-based registration

Published in: 6th Int. Conf. on Medical Image Computing and Computer-Aided
Intervention, MICCAI'2003, November 2003, Toronto, Canada.  Lecture Notes
in Computer Science 2879. Ellis, R. and Peters, T.  eds, Berlin: Springer,
2003. Vol. 1. p 351--358.